Virtualbox works for me if:

* The VM has System/MotherBoard/Chipset = PIIX3

* I boot with a grub command line of 'ATA=nolegacy ahci'

There seems to be a bug with the ATA drivers and Virtualbox's legacy IDE ports.

Quote:

Sounds like you've run into the 'APIC only interrupts' problem with AROS on pc-i386.

That error is due to reads timing out because the APIC interrupt specified in the PCI device's configuration isn't handled by AROS.

Someone should really fix that. It's getting harder and harder to find systems that correctly support legacy (IRQ < 16) interrupts.

Hi Jason,

i have to correct my last posting.
Firstly i have accidently tested today with the nightly from yesterday (2012-10-16)
With nightly from today (2012-10-17) AHCI is working in Virtualbox, if i use only SATA Controller.
If i use only IDE Controller it is also working, but enabling both kind of Controllers crashes AROS

Hi Jason,

yesterdays NB_ABI V0 works here on VirtualBox 4.1.22.

Attach file:

---

  AROS_ABI_V0_AHCI1.jpg (120.66 KB)

Its definitely closer than the last time I tried the nightlys, but still not usable on my machine. Using the Oct 17th nightlys, I'm also seeing is that an AFFS requester comes up with "read error -4 on block 0" and if I cancel that "read error -4 on block 8257032" and these two requesters repeat multiple times under either ABI with either native (IntelGMA) or VESA graphics. Eventually after enough cancels, the system starts to boot along further, ending in some strange state. In one case a boot shell that doesn't have any usable commands. Another time dead at the sponsored by imica splash screen.

In another case it came up with a shell that actually saw the live cdrom as volume ca0: when I tried the info command, so I typed execute s:startup-sequence and it came up to a wanderer screen after cancelling some cd-rom is write protected errors, and the workbench wants to close all windows message. Installaros failed to format/partition the drive that I had attached, so I couldn't install fresh from the nightly.

I guess I should try formatting and partitioning the hard drive for AROS outside the ahci driver and see if it is then visible when ahci is enabled.

Quote:

You also have the APIC interrupt delivery problem. Until AROS pc-i386 has APIC support, this cannot be resolved.

The problem is that the BIOS is setting the PCI interrupt line on the AHCI PCI device to an APIC interrupt (16 or higher).

You may want to try altering some BIOS settings related to interrupt routing (ie 'Legacy interrupts', or something like that), until AROS pc-i386 has APIC support.

If you can run Linux on this machine (or a *BSD), try running

# lspci -vvv | egrep '(^[0-9]|Interrupt)'

That should list all of the PCI devices and their interrupt routing.

I also have the APIC interrupt problem. That is damed shame :(

I don't have any BIOS options to change it either. What does it take to get APIC support ? Should we set up a bounty for it ?

Quote:

There (appears) to be APIC support in ABIv1 - it should be a Small Matter of Programming to backport it to ABIv0.

I'll play around with it. Maybe it won't be so hard as to require a bounty.

Ezrec: That would be much appreciated. Thank you for your hard work. I think that with AHCI more hardware will boot AROS :)